This class reads in raw images and publishes the location of the robot as well as a black and white occupancy. More...

#include <image_calibration.h>

Classes
struct	markerPoints
Public Member Functions
void	addClusters (vector< int > ImageXCoords, vector< int > ImageYCoords)
void	addMarkerPoint (int pointX, int pointY, int calibType)
void	applyBoundingBox (cv::Mat *image, cv::RotatedRect rect)
void	applyCalib ()
bool	applyCalibToPixel (int index, int calibType, cv_bridge::CvImagePtr image_msg, int camera)
string	convertInt (int number)
bool	extrapolateMarkers (vector< int > ImageXCoords, vector< int > ImageYCoords)
vector< cv::RotatedRect >	findBoundingBox (cv::Mat image, int xpos, int ypos, int *minDstIndex)
void	freerobotVector ()
void	getCalibration (int calibType, int calibR, int calibB, int calibG)
	imageCalibration ()
void	imageCallback1 (const sensor_msgs::ImageConstPtr &image_msg)
void	imageCallback2 (const sensor_msgs::ImageConstPtr &image_msg)
void	processImage ()
void	removeRobot (vector< int > xCoords, vector< int > yCoords, int xpos, int ypos)
void	resetForNextFrame ()
	~imageCalibration ()
Public Attributes
int	blurringFactor
int	calibAccuracy
calibTree *	calTree
int	colors [6][3]
int	count
ros::Subscriber	imgSub1
ros::Subscriber	imgSub2
int	lastFrameMidpointX
int	lastFrameMidpointY
cv_bridge::CvImagePtr	output_Img1
cv_bridge::CvImagePtr	output_Img2
cv::Mat	pastObstImg
ros::Publisher	pubImgCal
ros::Publisher	pubRobotPosition
cv_bridge::CvImagePtr	raw_Img1
cv_bridge::CvImagePtr	raw_Img2
vector< markerPoints >	robotMarkersVect
cv::Mat	roomBoundary
int	seq_num
ros::ServiceServer	service
bool	updatedImageFlag
bool	useSim
Private Attributes
string	filenames [5]

Detailed Description

This class reads in raw images and publishes the location of the robot as well as a black and white occupancy.

The imageCalibration class is used by launching the youbot_overhead_vision.launch file, or as part of the path_planning launch. It reads in raw images from two different overhead cameras topics (real or simulated in gazebo) and applies calibration files to these images to generate a single black and white image which can be used as an occupancy grid by the navigation stack. The robot heading and location are also found and published to a topic. The launch file contains parameters containing the location of the calibration files and the origin of the images (simulation or real cameras). There are 5 different types of calibrations. These are the floor, and four different robot markers (Left back, right back, left front, and right front). The files corresponding to these have points stored in them with rgb values. Calibration involves looking if each pixel in the images are close enough to any of the rgb values in these files, and changing the pixel to match the calibration color if they are. The floor file also contains the calibAccuracy variable, which determines how much is "close enough", and the blur variable, which determines what kind of blur to apply to the image before applying the calibration. The class uses the calibTree class to manage and parse calibration structures.

Definition at line 92 of file image_calibration.h.

Constructor & Destructor Documentation

imageCalibration::imageCalibration ( )

Constructor initializes parameters, publishers, and subscribers. This function also uses these parameters to get the calibration points from the calibration files, and generates the tree stored in calTree of all the calib objects for easier calibration point parsing.

Definition at line 16 of file image_calibration.cpp.

imageCalibration::~imageCalibration ( )

Definition at line 79 of file image_calibration.cpp.

Member Function Documentation

void imageCalibration::addClusters	(	vector< int > *	ImageXCoords,
		vector< int > *	ImageYCoords
	)

This function scans the robotMarkersVect vector and removes clusters that have less than MIN_MARKER_PIXELS pixels. Once thats done, the function determins which clusters are most likely the robot,and adds these to the structure holding the 4 marker positions.

Definition at line 399 of file image_calibration.cpp.

void imageCalibration::addMarkerPoint	(	int	pointX,
		int	pointY,
		int	calibType
	)

This function adds marker points to the markerPoints struct, determining the clusters A cluster is defined by a set of at least MIN_MARKER_PIXELS pixels less than MARKER_CLUSTER_RADIUS away from the center of the cluster.

Parameters:

pointX,:	The x position of the pixel in the image message
pointY,:	The y position of the pixel in the image message
calibType,:	The calibration type being applied to the pixel

Definition at line 360 of file image_calibration.cpp.

void imageCalibration::applyBoundingBox	(	cv::Mat *	image,
		cv::RotatedRect	rect
	)

Given a bounding box, removes everything in an image within that box

Parameters:

image,:	the image to apply the bounding box to
rect,:	the bounding box

Definition at line 852 of file image_calibration.cpp.

void imageCalibration::applyCalib ( )

This function applies a calibration file to the image. The tree generated in getCalibration() is used to parse the calibration point values in a rapid manner.

Definition at line 294 of file image_calibration.cpp.

bool imageCalibration::applyCalibToPixel	(	int	index,
		int	calibType,
		cv_bridge::CvImagePtr	image_msg,
		int	camera
	)

Helper function to applyCalib. Applys a specific calibration to a specific pixel.

Parameters:

index,:	marks the position of the pixel in the image data structure.
calibType,:	marks what type of calibration file is being applied to the pixel
image_msg,:	Contains the image being modified
camera,:	the camera whos image is currently being analyzed

Returns:: whether the calibType had a point which was close enough to the pixel values to calibrate the pixel to that color.

string imageCalibration::convertInt ( int number )

Converts an int to a string with that integer in it

Parameters:

number,the integer being put in the string.

Returns:: a string containing the number.

Definition at line 867 of file image_calibration.cpp.

bool imageCalibration::extrapolateMarkers	(	vector< int > *	ImageXCoords,
		vector< int > *	ImageYCoords
	)

This function fills in missing marker points depending on the points previously found on the robot. [1] = ROBOT_RIGHT_FRONT [2] = ROBOT_LEFT_BACK [3] = ROBOT_RIGHT_BACK NOT_SET (-1) means the marker values are empty Assumes only 2 points, but works for 3 points as well.

Parameters:

ImageXCoords,:	A vector containing x-coordinates of the marker centers.
ImageYCoords,:	A vector containing y-coordinates of the marker centers.

Definition at line 543 of file image_calibration.cpp.

vector< cv::RotatedRect > imageCalibration::findBoundingBox	(	cv::Mat	image,
		int	xpos,
		int	ypos,
		int *	minDstIndex
	)

Uses edge detection to determine a bounding box around the robot

Parameters:

image,:	the calibrated image from the overhead cameras
xpos,:	the x coordinate of the robot center
ypos,:	the y coordinate of the robot center

Returns:: a vector of all possible bounding boxes for the robot, with the most likely being the box at the minDstIndex

Definition at line 785 of file image_calibration.cpp.

void imageCalibration::freerobotVector ( )

This function cleans the robotMarkersVect structure

Definition at line 884 of file image_calibration.cpp.

void imageCalibration::getCalibration	(	int	calibType,
		int	calibR,
		int	calibB,
		int	calibG
	)

Function gets the calibration info, including the type of calibration that needs to be applied, and the colors that should be applied for each type.

Parameters:

	calibType: The type of calibration file applied
	CalibR: Red value for the color to use for the calibration
	CalibB: Blue value for the color to use for the calibration
	CalibG: Green value for the color to use for the calibration

Definition at line 84 of file image_calibration.cpp.

void imageCalibration::imageCallback1 ( const sensor_msgs::ImageConstPtr & image_msg )

Callback function for Camera 1 is called when a message is read on the /overhead_north/image_raw topic

Parameters:

image_msg,a structure containing the image acquired from the /overhead_north/image_raw topic

Definition at line 263 of file image_calibration.cpp.

void imageCalibration::imageCallback2 ( const sensor_msgs::ImageConstPtr & image_msg )

Callback function for Camera 2 is called when a message is read on the /overhead_south/image_raw topic

Parameters:

image_msg,a structure containing the image acquired from the /overhead_south/image_raw topic

Definition at line 278 of file image_calibration.cpp.

void imageCalibration::processImage ( )

Function processes overhead camera images. The two raw images (stored in raw_Img1 and raw_Img2) are combined into one image, eliminating the overlap between them. Then, the applyCalib() is used to apply the calibration points to the pixels in the image if applicable. Robot positions are printed to the /image_calibration/position_from_camera topic. Modified camera images are sent to the /image_calibration/image_cal topic.

Definition at line 136 of file image_calibration.cpp.

void imageCalibration::removeRobot	(	vector< int >	xCoords,
		vector< int >	yCoords,
		int	xpos,
		int	ypos
	)

Removes the robot from the image so that the resulting map can be used for path planning.

Parameters:

xCoords,:	the x coordinates of the robot marker points
yCoords,:	the y coordinates of the robot marker points.
xpos,:	the x coordinate of the robot center.
ypos,:	the y coordinate of the robot center.

Definition at line 663 of file image_calibration.cpp.

void imageCalibration::resetForNextFrame ( )

This function resets the structure containing the markers found in the last image in preparation for the next frame.

Definition at line 874 of file image_calibration.cpp.